This study will be part of a multicenter collaborative effort testing different strategies to achieve durable beta cell preservation in pre-diabetes/early T2D (RFA DK-10-013). The final intervention design and measurement endpoints will be harmonized within the consortium following selection of participating sites. We propose to study subjects with HbA1c 5.7-7.0% and 2 hour glucose >140 mg/dL following an oral glucose tolerance test. Subjects will initially be treated with insulin to achieve fasting glucose readingsof 90-100 mg/dL. Subjects will then be randomized to 6 month's treatment using either a strategy of beta cell support/unloading (using liraglutide plus pioglitazone) or one of beta cell rest/protection (using insulin Glargine plus the IL-1 receptor antagonist Anakinra). We will perform hyperglycemic clamps to measure first phase and second phase glucose-stimulated insulin secretion and use arginine stimulation to measure maximal secretory capacity as a surrogate of beta cell mass. We will also use oral glucose tolerance testing to measure integrated physiologic insulin responses including effects on the 2hr glucose reading. These tests will be performed at baseline (after relief of glucose toxicity) and after 3 and 6 months of randomized treatments. All subjects will transition to metformin monotherapy from months 6 to 24 (end of study), and undergo repeat measurements at 12 and 24 months. The treatment Response (magnitude of improvement in first-phase insulin secretion) and Durability (proportion of subjects maintaining improvements in first phase insulin secretion at 24 months greater than their own baseline at randomization) will be evaluated as coprimary endpoints. We will use baseline blood samples from the high and low responders from each of the above endpoints to undertake a biomarker discovery effort using proteomics. A high-throughput LC/MS/MS approach will be used for unbiased discovery and subsequent identification of circulating proteins that are robustly associated with these outcomes. Samples from local subjects will be used in the discovery phase, and samples from across the consortium will be used in the validation phase.